[Cmake] Re: Problems with Borland Generator?

[William A. Hoffman]

I think there is something wrong with your environment.
We are running several borland (5.5 and 5.6) dashboard builds from scratch each
night which build the entire cmake test suite and VTK. 
 From the error message, I can see that you have modified CMake because the output
should look like this:

Determining if the C compiler works passed with the following output:
MAKE Version 5.2  Copyright (c) 1987, 2000 Borland
Building object file testCCompiler.obj...
        bcc32  at MAKE0001. at  at  at 
Borland C++ 5.5.1 for Win32 Copyright (c) 1993, 2000 Borland
c:\hoffman\CMake_borland_dashboard\Tests\Simple\CMakeTmp\testCCompiler.c:
Building executable c:\hoffman\CMake_borland_dashboard\Tests\Simple\CMakeTmp\cmT
ryCompileExec.exe...
        bcc32  at MAKE0000. at  at  at 
Borland C++ 5.5.1 for Win32 Copyright (c) 1993, 2000 Borland
Turbo Incremental Link 5.00 Copyright (c) 1997, 2000 Borland


In your imate, the command looks like this:
bcc32 testccompiler.obj -o ....
Which is not using the file redirection, and is missing many of
the flags that are used by default with the borland compiler.

Please try again with an unmodified CMake.


I am not really sure what you are talking about with the 
TARGET__SRC_OBJS stuff??   

It actually says this:

simple_SRC_OBJS = \
simple.obj 

simple.obj: c:\hoffman\CMakeDashboardSrc\Tests\Simple\simple.cxx
         at echo Building object file simple.obj...
        bcc32   at &&|
-DWIN32 -P     -w- -whid -waus -wpar -tWM -Od -v $(INCLUDE_FLAGS)   -osimple.obj -c c:\hoffman\CMakeDashboardSrc\Tests\Simple\simple.cxx
|


c:\hoffman\CMake_borland_dashboard\Tests\Simple\simple.exe: $(simple_SRC_OBJS) $(simple_EXTERNAL_OBJS) $(simple_DEPEND_LIBS)
         at echo Building executable c:\hoffman\CMake_borland_dashboard\Tests\Simple\simple.exe...
        bcc32  at &&|
  -tWM -lS:10000000 -lSc:10000000  -v -tWC  -ec:\hoffman\CMake_borland_dashboard\Tests\Simple\simple.exe   -w- -whid -waus -wpar -tWM    -Lc:\hoffman\CMake_borland_dashboard\Tests\Simple simpleLib.lib import32.lib  $(simple_SRC_OBJS) $(simple_EXTERNAL_OBJS)  
|

No full path is used for the .obj file because it is in the current directory.
I see nothing wrong with using the full path for the .cxx and no full path for a .obj file.

However, perhaps your modifications have broken something else....


-Bill

At 06:47 PM 4/19/2004, John Biddiscombe wrote:
>Resending with smaller jpg...
>
>----- Original Message ----- 
>From: "John Biddiscombe" <jbiddiscombe at skippingmouse.co.uk>
>To: <cmake at www.cmake.org>
>Sent: Monday, April 19, 2004 11:43 PM
>Subject: Problems with Borland Generator?
>
>
>> I've been having some trouble with my custom generator which is a subclass
>> of the Borland Makefiles generator, so I did a new cvs Checkout of CMake
>and
>> tried building a project with it (using the supplied Borland Generator)
>>
>> The attached jpg shows the initial error I get with it.
>>
>> I note from looking at various makefiles that the borland generator (and I
>> guess all others descended from the unix generator) creates makefiles with
>>
>> Target_SRC_OBJS =
>> aFile.cpp
>> anotherfile.cpp
>> etc etc etc
>>
>> Rules are specified as
>>
>> C:\MyStuff\Dir\aFile.cpp : Stuff
>>
>> C:\MyStuff\Dir\anotherFile.cpp : Stuff
>>
>> Now The borland make won't use the correct rules for these because the
>Rule
>> refers to a full path, whereas the dependency is only a relative path.
>This
>> means that the default compiler options are always used no matter what you
>> set in your various options. (this has little bearing on the attached jpg,
>> but there's obviously a path problem somewhere there too)
>>
>> Could someone please try a completely clean build of a Borland project
>with
>> a recent checkout of CVS (today/yesterday etc) and tell me if something
>has
>> gone horribly wrong at my end, of if there is indeed a bug. I think
>there's
>> a bug, but then I've got so many build trees lying around that I've surely
>> messed up somewhere.
>>
>> Many thanks and apologies if this turns out to be a glitch at my end.
>>
>> JB
>>